Department of Statistics Seminar
North Carolina State University

presents

Dr. Alan E. Gelfand

University of Connecticut

"Inference for Spatially Misaligned Data"

ABSTRACT

In attempting to establish relationships between spatial variables one often encounters misaligned data layers. For instance, data layers may arise as point-referenced data, i.e., observations at points in a region or on areal units, i.e., observations on partitions of the region. In the former case, different layers may be observed at different sets of points. In the latter case, different layers may be on different partitions. In some datasets we may have both areal and point-referenced layers. In attempting to reconcile such misalignment for the purposes of regression, interpolation and prediction, one finds a variety of ad hoc methods in the literature. We propose fully model-based approaches to address these issues. In particular, we employ hierarchical models and provide full inference. We adopt a Bayesian approach, avoiding likelihood asymptotics which are typically inappropriate. We illustrate with several environmental and ecological examples.

Friday, April 14, 2000

3:35 - 4:35 pm

206 Cox Hall

Refreshments will be served on the second floor of Dabney Hall (left of Room 222) at 3:00 pm.